Databending an image with an audio editor like Audacity achieves very different results than with Text or Hex editors. You can rearrange the waveform with cut and paste or use effects much like photoshop filters. This article will focus on using audio effects.
Databending JPGs doesn’t work very well, they tend to break on export. This image was converted to .BMP then opened in Audacity as Raw Data.
I’ve found that the best import settings for .BMP files is Encoding: A-Law; Byte Order: Little-Endian; Channels: 1 Channel.
The file opens and shows an audio interpretation of the data. It’s not a nice sound, I’ll speak more on that in another article.
In the waveform we can identify a short header section then pixels from top left to bottom right. Knowing this we can avoid editing the header (or else file breaks) and edit image pixels horizontally. Manipulating the image in the vertical axis would be extremely tedious. You’d have to figure out a specific pixel location in samples.
With this image’s waveform I was able to identify approximately where the face was between the text. I made a time selection and experimented with effects. It was interesting that the results were somewhat related to the audio processing.
To export the image set the file type to “Other uncompressed” and choose RAW (headerless) and A-Law for encoding in options. You have to manually enter the .bmp extension. Ignore the error message.
Pretty cool IMO. I like the echo, phaser, and EQ most. I like that you can almost control it but the results are still a little unexpected.